Anamorphic Cinemascope in Native 1920x810 MJPEG Mode

texasmfp said:
Hi: so I have a real set of 35mm anamorphics (LOMO roundfronts). What is my best option and settings for ending up with 2:35:1 video? Thanks
Click to expand...
I measured the squeeze factor of my Lomo Square Front adapter as about 1.85x which is not a standard I've seen anywhere else. I believe Lomo Round Fronts have a 2x squeeze factor, but I'd recommend testing it yourself.

At this time, I think the most flexible MJPEG setup for anamorphic recording is the 1920x810 frame size I developed for 1.33x adapters like the LA7200 and Century Optics. When using this mode with other types of anamorphics, multiply the squeeze factor by 1440 to calculate the correct frame width to stretch the video to in post-production.
 
johnnym said:
If money is no problem, go with a good Iscorama on ebay. Be prepared to pay a lot of cash though, in the region of 1000-2000$. Those are top of the line of the anamorphic adapters. It is rumoured Isco will release a new adapter in the future, but last time i checked it hadn't been released. Check http://www.isco.eu/index.php?id=taking-lenses
I think the Century Optics is really good (by the way they're part of Isco now) , but i'd like to see LPowell use it on a 50mm and see how much blur remains on the sides.

Also, you can always buy Anamorphic lenses, but they are rarely found 2nd hand, except for the Russian Lomo lenses. As i understand it, all films shot on anamorphic use lenses hired from Panavision. You have some other brands of anamorphic lenses, like Hawk, etc. There are places online where they can be bought, but always at prices +2000$.
Click to expand...

Wow, thanks for all the info!

Does it matter which Iscorama? Or any one that is 1.33x? 54s or 36s? 36 is 1.5x?

And there is no barrel distortion at all with this one?

And this might seem like a dumb question, but would shooting with anamorphic lenses provide better quality since there is no adapter at all?
 
Hmm, my GH13 is producing some corrupted MPEG files. I will try to post them. Apart from infrequent freezes (1 for an hour of shooting today) I am getting choppy MPEGs (as if my shutter speed is really low). I will try to find a tool to analyze the corrupted clips and will post again.
I am running the vanilla MPEG patch now (took away any AVCHD mods, left the parts relevant for MPEG only)...
 
Bfilms said:
Wow, thanks for all the info!

Does it matter which Iscorama? Or any one that is 1.33x? 54s or 36s? 36 is 1.5x?

And there is no barrel distortion at all with this one?

And this might seem like a dumb question, but would shooting with anamorphic lenses provide better quality since there is no adapter at all?
Click to expand...


Here's a test without and with the Iscorama 36 on a Nikon 50mm f/1.8, one of the sharpest lenses available. http://www.cinema5d.com/viewtopic.php?p=60952
I'm no expert but i think it performs rather well. The author says it does have barrel distortion. In my opinion the LA7200 exhibits more of it (look at some portraits filmed), but search for footage on Vimeo and find your own conclusion.

I can't tell you anything on the anamorphic lenses. I have seen footage of the Lomos and i don't like it, but that's just my taste. Search on Vimeo for Lomo lenses and you'll know whether you do.

Here's some tests of a 2x Iscorama and a 2x Kowa http://vimeo.com/16350276 I own a Kowa 16H and it's a really good anamorphic lens but focusing is a pain compared to the Isco. On the other hand, you ocan buy 10 of these Kowas for 1 Isco on ebay. Prices for Iscoramas are really quite insane at the moment. (Although the new Iscos will probably be double the price you pay for a used one now)

See here for much more footage : http://vimeo.com/groups/anamorphic/
 
Last edited:
lpowell said:
Sorry, but the files at your link don't seem to download or play.
Click to expand...

Strange. I am uploading them to rapidshare and depositfiles and will update the links as soon as they are uploaded.

Update: Here are the links

--Rapidshare--
http://rapidshare.com/files/428812293/1.MOV
http://rapidshare.com/files/428813193/2.MOV
http://rapidshare.com/files/428816439/3.MOV

--Depositfiles--
http://depositfiles.com/files/ixjxejsbp
http://depositfiles.com/files/c8cnrq6qn
http://depositfiles.com/files/44wisdijh

I hope those links work. And thanks for taking the time to check them out.
 
Last edited:
Irfan,
great work with the tests. Interesting that the FCP method seems lighter on the various tests you did. It looks good though and I just loaded the patch on my GF1 and will try it out
 
As Lpowell asked me.. I will post my latest test results again.
Now I can tell that MJPEG is not just upscaled.. It contains more information's then AVCHD 720p. I upscale AVCHD 720p to 1080p and compare it to MJPEG 1080p converted from Anamorphic 810.
All footage was shoted with GF1 in Smooth Film mode with -2 settings.
Result are below:

AVCHD 720p upscaled to 1080p:


MJPEG Anamorphic patch 810 to 1080p:





TIP:If you are using Quicktime you can easly upsample your Anamorphic footage to any dimensions you need.
1. Open your footage in quicktime (by default on Mac)
2. CMD+J or in menu Movie properties
3. Chose Video track then go to Visual setting tab
4. Input your desired dimensions
5. And export it in desired video format.
 
Last edited:
Thanks for sharing. I think the difference is quite big. If you can, i'd be interested to see the difference in detail between AVCHD 1080p and the anamorphic MJPEG
 
Unfortunately I am not able to shoot AVCHD 1080p since I have GF1 only. Maybe someone else could try that.
 
Thanks for making your results available, Irfan. I think these screenshot comparisons between MJPEG 1280x720 and 1920x810 frames upscaled to 1080p have disproven the assumption that custom MJPEG modes are scaled up from 1280x720 frames.

Early attempts to use custom MJPEG modes focused on 1080p height while short-changing horizontal resolution. My recent testing led me to conclude that achieving high-bitrate 1920-wide horizontal resolution produces visibly superior results. In addition, with the use of a widescreen 2.37:1 aspect ratio, only 810-high vertical resolution is needed.

All MJPEG modes must use scaling to produce an HD video frame from the 12-megapixel sensor. One major clue to what's going on under the hood is the encoder's ability to produce both 16:9 and 4:3 aspect ratios in its built-in MJPEG video modes. This strongly suggests that the encoder uses a variable scaling algorithm that can natively produce any desired aspect ratio, rather than first scaling to 1280x720 and then scaling to custom modes from there.
 
Some footage i shot in VGA mode with the Elmoscope II.


Very strange is the pixelation i noticed in AE, when the composition is set to anamorphic.
Because of this i set my comp to use square pixels at a resolution of 1280x480, instead of 640x480 (with anamorphic 2:1). Examples below

AE-Comp-Ana.jpg This is with the comp set to 2:1 anamorphic
AE-CompSquare.jpg This is with the comp set to square pixels (but twice as wide)
 
Last edited:
Here is a small compilation of test footage I took with the anamorphic patch and my 20mm pancake (everything except the timelapse in the middle).
Scaled to 720p before uploading to Vimeo. Downloading the original file is recommended.

http://vimeo.com/16702569
 
I'm trying to combine this with the 50MBS 3-GOP patch. AVCHD recording works fine but I keep on getting data card errors (Transcend class 6) when recording MJPEG. Does this mean I'll need to reduce the E values for the MJPEG compression settings?
 
You must log in or register to reply here.
Back
Top